Impact of epidermal growth factor receptor expression on survival and pattern of relapse in patients with advanced head and neck carcinoma.

Abstract

A correlative study was performed to address the impact of epidermal growth factor receptor (EGFR) overexpression on survival and pattern of failure in patients with advanced head and neck squamous cell carcinomas (HNSCCs) enrolled in a Phase III trial and randomized to receive conventional radiotherapy. The study population comprised 155 of 268 (58%) randomized patients with sufficient pretreatment biopsy specimens for immunohistochemical assay. The specimens were dewaxed and incubated after standard preparation with mouse monoclonal antibodies recognizing the extracellular domain of the EGFR molecule. The catalyzed product was visualized with 3,3'-diaminobenzidine Chromogen Kit and lightly counterstained with Mayer's hematoxylin. Quantitative EGFR immunohistochemistry (IHC) was done with SAMBA 4000 Cell Image Analysis System, without knowledge of the clinical outcome, to yield mean absorbance (MOD), staining index (SI), and quick score (QS). These EGFR IHC parameters were correlated with the T stage, N stage, combined stage grouping, and recursive partitioning analysis classes. Subsequently, the EGFR parameters were correlated with the outcome end points, i.e., overall survival (OS), disease-free survival (DFS), local-regional (LR) relapse, and distant metastasis rates. We found that HNSCCs exhibited a wide variation in EGFR expression (MOD, 0.2-66.0; SI, 0.3-97.0; QS, 0.01-69.9) with a relatively strong but nonlinear correlation between MOD and SI (r = 0.79). There was no correlation between EGFR expression and T stage, N stage, stage grouping, and recursive partitioning analysis classes (r = -0.07 to 0.17). The OS and DFS rates of patients with high EGFR-expressing HNSCCs (>median MOD) were highly significantly lower (P = 0.0006 and P = 0.0016, respectively) and the LR relapse rate was highly significantly higher (P = 0.0031) compared with those of patients with low EGFR-expressing HNSCCs. However, there was no difference in the distant metastasis rate between the two groups (P = 0.96). Significant correlations, although somewhat less robust than MOD, were also observed between SI and QS and the OS, DFS, and LR relapse rates. Multivariate analysis showed that EGFR expression was an independent determinant of survival and a robust independent predictor of LR relapse. In summary, this correlative study in a large series of patients revealed that EGFR expression, which varied considerably among HNSCCs, was a strong independent prognostic indicator for OS and DFS and a robust predictor for LR relapse but not for distant metastasis. The data suggest that EGFR IHC should be considered for selecting patients for more aggressive combined therapies or enrollment into trials targeting EGFR signaling pathways.